METHODS & TECHNIQUES

Establishing simple image-based methods and a cost-effectiveinstrument for toxicity assessment on circadian rhythmdysregulation in fishGilbert Audira1,2,*, Bonifasius Putera Sampurna2,*, Stevhen Juniardi2, Sung-Tzu Liang2, Yu-Heng Lai3,Liwen Han4,‡ and Chung-Der Hsiao1,2,5,6,‡

ABSTRACTAnalysis of circadian rhythm behavior alteration in fish for toxicityassessment usually requires expensive commercial equipmentand laborious and complicated tweaking. Here, we report a simplesetup that consists of a custom-made light box equipped with whiteand 940 nm light-emitting diode (LED) light strips as light sources,where the locomotion activities of zebrafish or catfish are capturedusing an infrared-sensitive coupled charged device (CCD). Thewholesetup was housed in a temperature-controlled incubator to isolateexternal noise and to maintain consistent experimental conditions.The video recording and light triggering were synchronized usingTotal Recorder, a recording scheduling software. By using the setupmentioned above and open source software such as ImageJ oridTracker, the locomotion activities of diurnal (e.g. zebrafish) andnocturnal (e.g. catfish) fish during day and night cycles can bequantitatively analyzed.We used simple image-basedmethods and acost-effective instrument to assess the circadian rhythm of multiplefish species, as well as other parameters such as age, ambienttemperature and chemical toxicology with high precision andreproducibility. In conclusion, the instrument setting and analysismethods established in this study provide a reliable and easy entrypoint for toxicity assessment on circadian rhythmdysregulation in fish.

KEY WORDS: Circadian rhythm, IdTracker, ImageJ, Locomotion

INTRODUCTIONCircadian rhythms are based on intracellular time-tracking systemsthat play a central role in adapting the physiology and behaviorof living organisms to anticipate daily environmental changes(Lahiri et al., 2005). The principal cue that influences circadianrhythmicity is light (Hirayama et al., 2005). Circadian rhythmsare driven by endogenous circadian clocks that are comprised of

self-sustained oscillators that drive rhythms within a period near24 h in the absence of external timing cues. Circadian clocksinvolve clock genes that interact to produce a molecular oscillatoradjusting output clock-controlled genes. Thus far, some clock geneshave been identified in vertebrates and most of the studies on themolecular mechanisms of circadian rhythm generation have comefrom research of clock genes that were originally identified bymutations, which altered or abolished circadian rhythms (Delaunayet al., 2000). To date, many circadian rhythm mutants have beendiscovered in Arabidopsis, Drosophila, Neurospora andSynechococcus. However, only a handful of circadian genes havebeen discovered in vertebrates.

In the past few years, the use of zebrafish (Danio rerio) has beenexpanding from its traditional user base to studying variousbiomedically relevant aspects of biology, including behaviorand physiology. The rapid accumulation of zebrafish genetic andgenomic information also makes zebrafish potentially usefulvertebrates for mutational analysis of vertebrate circadianrhythmicity, especially in circadian gene regulation and functionsof the pineal organ (Gamse et al., 2002). The zebrafish circadiansystem shares some similarities with that of Drosophila, and theirtissues contain endogenous circadian oscillators that are lightresponsive and entrainable to light–dark (LD) cycles in vitro (Tamaiet al., 2007). Even though the interest in zebrafish as a vertebratemodel organism does not focus on the circadian clock, severalaspects of their basic biology make them inherently ideal for thisarea of study. First of all, their embryos are completely transparentand develop rapidly in an egg shell or chorion, so the wholedevelopmental process can be observed non-invasively under themicroscope. Secondly, their reproduction is relatively simple andthey can be raised at high density at low cost. In addition, adultzebrafish are hardy, small (2–3 cm long) and reach sexual maturityafter only 2–3 months. Another advantage is that zebrafish are idealfor large-scale forward and reversed genetic screening (Vatine et al.,2011). Because of their potential usefulness for circadianrhythmicity research, efficient methods for analysis of circadianrhythm phenotypes in zebrafish are needed (Cahill et al., 1998).

Maintained circadian locomotor activity rhythms of animals inconstant conditions have been utilized to study the functions ofsome clock genes and to screen for circadian clock mutants (Antochet al., 1997; King et al., 1997; Konopka and Benzer, 1971; Ralphand Menaker, 1988). These locomotor rhythms have also been usedcomprehensively to analyze the physiological organization ofmammalian, reptilian and avian circadian systems. However, fewstudies of teleost locomotor rhythmicity in constant conditionscan be found in the literature, even though current available datareveal considerable variability in the strength and pattern of teleostlocomotor activity rhythms. To date, there have been few reports ofReceived 22 January 2019; Accepted 28 May 2019

1Department of Chemistry, Chung Yuan Christian University, Chung-Li 32023,Taiwan. 2Department of Bioscience Technology, Chung Yuan Christian University,Chung-Li 32023, Taiwan. 3Department of Chemistry, Chinese Culture University,Taipei 11114, Taiwan. 4Biology Institute, Qilu University of Technology (ShandongAcademy of Sciences), Jinan City, Shandong, China. 5Center of Nanotechnology,Chung Yuan Christian University, Chung-Li 32023, Taiwan. 6Center of BiomedicalTechnology, Chung Yuan Christian University, Chung-Li 32023, Taiwan.

*These authors contributed equally to this work

‡Authors for correspondence (hanliwen08@126.com; cdhsiao@cycu.edu.tw)

C.-D.H., 0000-0002-6398-8672

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2019. Published by The Company of Biologists Ltd | Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

circadian behavioral rhythmicity in zebrafish (Hurd et al., 1998).Several techniques for the characterization of behavioral circadianrhythmicity in zebrafish have been developed. However, most of theassessments were conducted in zebrafish larvae and the activityrecords were analyzed manually (Cahill et al., 1998; Hurd andCahill, 2002). Moreover, using either larvae or adult zebrafish, theobtained data cannot describe their precise spatial information(Hurd et al., 1998; Zhdanova et al., 2008). In addition, we alsotested our method in Indian walking catfish (Clarias batrachus).These important commercial species of fish were used becausethey are nocturnal animals and their circadian rhythm of locomotoractivity can help to validate the usability of our current method.However, even though the catfish is a nocturnal animal, it exhibitsLD-entrainment of its locomotor activity that free-runs underconstant conditions. It appears that the overt circadian rhythm oflocomotion in this fish is underlain by multiple oscillators (Ramtekeet al., 2009). Therefore, acclimatization for these fish to the currentphoto-regime condition is mandatory before the experiment.Overall, in this study, we aim to develop an automatic andreliable circadian activity measurement method and relatedinstrument for zebrafish and other fish species (Fig. 1).

RESULTSMeasurement of the circadian rhythm in zebrafishand catfishTo explore the potential utility of these two protocols, we validatedthe method by using diurnal (zebrafish, D. rerio) and nocturnal(catfish, C. batrachus) fish species with distinct locomotionactivities during the light and dark cycles. For zebrafish, wefound robust swimming activity during the light cycle and a sleep-like behavior during the dark cycle (Fig. 2B and Movie 1). This wasexemplified by higher than average speed and rapid movement–time ratio during the light cycle compared to the dark cycle(Fig. S1D,E). Furthermore, this behavior was also demonstratedby the low level of meandering and freezing time–movement ratioexhibited by zebrafish during the light cycle (Fig. S1F,G). On theother hand, catfish were active in the dark cycle and showed a sleep-like behavior during the light cycle (Fig. 2C and Movie 2). Thisbehavior was indicated by a lower than average speed and rapid

movement–time ratio during the light cycle compared to the darkcycle (Fig. S1H,I). In addition, high levels of freezing time–movement ratio and meandering were observed in this experiment(Fig. S1J,K). This result is in line with the literature showing mostcatfish species have nocturnal behaviors (Vera et al., 2011; Ramtekeet al., 2009). The locomotor activity of catfish is entrained by light–dark cycles. Catfish activity increased during the dark periods,indicating that catfish are nocturnal. However, zebrafish locomotoractivity intensified during the light periods, showing that zebrafishare diurnal (Kasai et al., 2009). For both fish species with distinctcircadian rhythms, we cross-examined the locomotion activitycurves acquired by idTracker- and ImageJ-based methods. Afterdata normalization, an unpaired t-test with Welch’s correction wasused to analyze the differences between these data. Later, we foundthat both data were paralleled and correlated with each other,indicating that both methods can be used to measure fish circadianrhythm activities with high reliability and reproducibility.Furthermore, from the r-values that we obtained from Spearman’snonparametric correlation test, we concluded that the two variables(ImageJ and idTracker results) tend to increase or decrease together,or in other words, they correlate with each other, with high r-valuessupported (Fig. 2D,E for zebrafish and Fig. 2F,G for catfish).

Measurement of the circadian rhythm in zebrafish withdifferent age and temperatureAging is the accumulation of deleterious changes over multiplelevels of biological organization resulting in a decrease in whole-organism functionality over time (Gilbert et al., 2014). The resultsof multiple studies in humans and animal models suggest thataging alters the circadian clock (Zhdanova et al., 2008). In orderto study the age effect in zebrafish circadian rhythm, we measuredthe circadian rhythm locomotion activities for 6-month-old (n=28)and 16-month-old (n=18) zebrafish. We found that the olderzebrafish showed higher locomotion in the dark cycle and lowerlocomotion in the light cycle compared to younger zebrafish(6-month-old) (Fig. 3A, Fig. S1A and Movie 2). Thisphenomenon is supported by a low level of average speed and ahigh level of meandering in the light cycle (Fig. 3C,E) and highlevels of both average speed and average angular velocity in the

Fig. 1. Overview of the custom-made light box chamber to detect the circadian rhythm in fish. (A) Schematic of the fish circadian rhythm chamber,which consisted of an IR camera (IR CCD), an LED light box and a temperature control incubator. (B,C) Example snapshots of fish position in the water tankduring the (B) light (day) and (C) dark (night) cycles. (D) LED light source arrangement for the fish circadian rhythm chamber.

2

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

dark cycle (Fig. 3F,G). Meanwhile, there was no difference inaverage angular velocity during the light cycle and meanderingduring the dark cycle between the adult fish and the old fish(Fig. 3D,H). This result is in agreement with a previouslypublished study showing that zebrafish aging is associated withsignificant changes in rhythms of activity, sleep and melatoninproduction (Zhdanova et al., 2008). In the dark cycle, reducedsleep-like behavior was exhibited by the older tested zebrafish.Sleep-like behavior in zebrafish is classified as prolonged periodsof inactivity (over 5 s). During this period, zebrafish may staymotionless or occasionally exhibit very slowmovement. Melatonin, ahormone produced in the pineal gland, induces sleep in somecontexts and plays a key role for mediating the circadian processbecause clock regulates its production (Gandhi et al., 2015).Melatonin promotes sleep in diurnal animals, including humansand zebrafish, and its production in the middle of the dark phasegradually declines with age. Therefore, reduced hours of sleep time isfound in older people, as well as an incrementing number ofawakenings and a reduction in the deep-sleep stage of non-REM.Several studies also report that older humans have difficultymaintaining sleep on a regular basis (Duffy et al., 2015; Dijk et al.,2001; Ehlers and Kupfer, 1989; Webb, 1982).Temperature is considered the ‘abiotic master factor’ among

physical factors that can influence behavior, physiology and

distribution of aquatic organisms in the aquatic environment. Dueto its importance, thermal tolerance has been extensively studied infishes (López-Olmeda and Sánchez-Vázquez, 2011). For potentialtemperature effects in the circadian rhythm of zebrafish, wemeasured the circadian rhythm locomotion activities of 6-month-old zebrafish in ambient temperatures of either 18°C (n=13), 25°C(n=28) or 30°C (n=18) and found there were significant locomotoractivity differences in their circadian rhythm (Fig. 3B, Fig. S2A,Band Movie 3). When the temperature was changed from 25°C to18°C, the activity of zebrafish during the light and dark cyclesdecreased. In the light cycle, this behavior was indicated by lowerthan average speed and angular velocity, and a high level ofmeandering (Fig. 3I–K). Meanwhile, lower than average speed andhigher meandering than the control fish in the dark cycle alsosupported this phenomenon (Fig. 3L,N). In addition, there was nodifference in average angular velocity in the dark cycle between thelow and normal temperature conditions observed in the testedzebrafish (Fig. 3M). In line with a previous study by Lopez et al.,total activity of zebrafish during the day decreased to about 30%when the constant temperature was reduced (López-Olmeda et al.,2006). Furthermore, high temperature also can alter zebrafishcircadian rhythm behavior, especially in the light cycle. Thisalteration, demonstrated by higher than average speed and low levelof meandering, indicates hyperactivity behavior (Fig. 3I,K).

Fig. 2. Comparison of circadian rhythm locomotion activities of zebrafish and catfish measured by the idTracker or ImageJ methods. (A) Schematicof the analysis pipeline for fish locomotion measurement in this study. (B,C) The circadian rhythm of diurnal fish species (B, zebrafish) and nocturnal fishspecies (C, catfish) identified by either idTracker (blue) or ImageJ (red). The data are expressed as the means. (D,E) Comparison of normalized percentagesbetween idTracker and ImageJ results after data normalization in total zebrafish circadian activity in (D) day cycle and (E) night cycle with r-value (highlightedin blue) from Spearman’s nonparametric correlation test. (F,G) Comparison of normalized percentages between idTracker and ImageJ results after datanormalization in total catfish circadian activity in (F) day cycle and (G) night cycle with r-value (highlighted in blue) from Spearman’s nonparametriccorrelation test. The data are expressed as the means±s.e.m. (zebrafish, n=18; catfish, n=6) and were analyzed by unpaired t-test with Welch’s correction.

3

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

Meanwhile, the same level of average angular velocity betweennormal-and high-temperature zebrafish was found (Fig. 3J). Inaddition, a higher temperature also slightly altered zebrafishbehavior in the dark cycle, demonstrated by higher than averageangular velocity and meandering compared to the normaltemperature as control (Fig. 3M,N and Movie 3). No significantdifference was found in average speeds of fish during the dark cycle(Fig. 3L). These findings are in agreement with those of a previouswork by McClelland et al., which reported an increase in themaximum sustained swimming speed in zebrafish reared at 28°C.Lopez et al. also found that zebrafish showed higher than averagedaily activity when they were maintained at high temperature(López-Olmeda et al., 2006; McClelland et al., 2006). Supportingthese phenomena, another study already found that temperature hada significant effect on the proportion of animals that exhibitedstatistically significant rhythmicity. It could be one of theenvironmental conditions that affects coupling among oscillators;temperature could modulate factors that are not related to circadianrhythms. For example, low or high temperatures could suppress orinduce activity and mask the endogenous rhythm of locomotoractivity (Hurd et al., 1998). Later, circadian rhythm patterns for age-and temperature-effect groups were analyzed by idTracker andImageJ and were combined and compared (Fig. S2A,B). In order tocross-examine the locomotion activity data acquired by these twomethods, data normalization and statistical analysis were conductedand, later, we found that all data were paralleled and correlated witheach other. This finding is also supported by r-values that weobtained from Spearman’s correlation test, which concluded that

these two variables (ImageJ and idTracker results) have a tendencyto increase or decrease together (Fig. S1B,C for age effect andFig. S2C–F for temperature effect).

Measurement of the circadian rhythm dysregulation inzebrafish after ethanol exposureAlcohol intake and chronic alcoholism in humans are associatedwith disruptions of sleep–wake cycles and other daily biologicalrhythms. Similar disruptions also have been seen in experimentalanimals treated with chronic ethanol (EtOH) (Rosenwasser et al.,2005). To explore the effects of acute and chronic EtOH ingestionon circadian activity in the zebrafish, we measured the circadianrhythm locomotion activities of 6-month-old zebrafish exposed toacute (∼30 min) and chronic (∼7 days) 0.1% EtOH. Alcoholmixed in the water of the test tank was absorbed by the bloodvessels of the skin and the gills of the fish, therefore blood-alcohollevels reached equilibrium with the external alcohol concentrationquickly (Gerlai et al., 2000). We found that acute and chronicadministration of low concentrations of EtOH altered circadianrhythm activity in both light and dark cycles (Fig. 4A, Fig. S3A,Band Movie 4). In the acute treatment of EtOH, this alteration wasindicated by hyperactive behavior, demonstrated by higher thanaverage speeds than control fish in both cycles (Fig. 4B,E).Furthermore, lower than average angular velocity and meanderingin acutely-treated fish in the light cycle showed that the fish wereinfrequently changing their swimming direction (Fig. 4C,D). Inaddition, sleep disruption caused by acute administration of EtOHwas also observed in this experiment, exemplified by similar

Fig. 3. Effects of different ages and temperatures on circadian rhythm locomotion activities of zebrafish. (A) Comparison of the circadian rhythmlocomotion activities between 6-month-old (black) as a control and 16-month-old (red) zebrafish. (B) Comparison of the circadian rhythm locomotion activitiesfor zebrafish acclimated to either ambient temperature at 18 (blue), 25 (black) or 30°C (red). (C–H) Comparison of average speed (C,F), average angularvelocity (D,G) and meandering (E,H) of zebrafish between the old fish and control fish during the light and dark cycles. The data are expressed as themeans±s.e.m. (control, n=28; old fish, n=18) and were analyzed by unpaired t-test. (I–N) Comparison of average speed (I,L), average angular velocity (J,M)and meandering (K,N) of zebrafish when they incubated in 18 (blue), 25 (black) or 30°C (red) during the light and dark cycle. The data are expressed as themeans±s.e.m. (18°C, n =13; 25°C, n=28; 30°C, n=18) and analyzed by Dunnett’s multiple comparisons test. All of the multiple comparisons tests comparedthe mean of each column with the mean of a control column. Significance difference was defined as *P<0.05, ***P<0.001, ****P<0.0001.

4

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

patterns of swimming direction between the acutely-treated fish inthe dark cycle and control fish in the light cycle (Fig. 4F,G).Chronic exposure of 0.1% EtOH, on the other hand, reduced fishactivity during the light cycle. This is hypoactivity behavior,

indicated by low levels of average speed and high levels ofmeandering (Fig. 4B,D). Meanwhile, there was no significantdifference in average angular velocity between the controls andchronically EtOH-treated fish during the light cycle (Fig. 4C).

Fig. 4. The circadian rhythm locomotion activities of zebrafish exposed to acute and chronic 0.1% EtOH. (A) Comparison of the effects of acute(∼30 min, blue) and chronic (∼7 days, red) treatments with 0.1% EtOH on zebrafish circadian rhythm locomotion activities. (B–G) Comparison of averagespeed (B,E), average angular velocity (C,F) and meandering (D,G) between acutely-treated, chronically-treated and control fish during the light and darkcycle. The data are expressed as the means±s.e.m. (control, n=28; acute, n=18; chronic, n=18) and were analyzed by Dunnett’s multiple comparisons test.All of the multiple comparisons tests compared the mean of each column with the mean of a control column. Significance difference was defined as *P<0.05,**P<0.01, ***P<0.001, ****P<0.0001.

5

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

Similar to acute treatment of EtOH, sleep impairment was alsofound during the dark cycle in the chronically treated-fish. Thisphenomenon was indicated by higher than average speed andangular velocity in EtOH-treated fish compared to the control fish(Fig. 4E,F). In addition, there was no significant difference in themeandering for the chronically EtOH-treated fish during the darkcycle (Fig. 4G). These results are in agreement with previousstudies about the effect of acute EtOH administration to zebrafishactivity; at lower concentrations (e.g. 0.25% and 0.50%), EtOHwill increase zebrafish locomotion activity and it will depress theiractivity at a higher dose (1.00%) (Gerlai et al., 2000). Anotherstudy by Lockwood et al. in larval zebrafish showed that theyexhibit acute sensitivity to EtOH in a dose- and time-dependentmanner. Firstly, they initially become hyperactive and as EtOHaccumulates, they become hypoactive and sedated, which issimilar to what has been observed in humans and other animalmodels (Lockwood et al., 2004). These prior findings may explainthe hyperactivity behavior in the acute EtOH-treated fish andhypoactivity behavior in the chronic EtOH-treated fish during thelight cycle. However, there is no study about the effect ofEtOH treatment on circadian rhythm activity in adult zebrafishpublished until now. The rat study by Rosenwasser et al.showed that both EtOH withdrawal and EtOH intake alter theperiod and amplitude of free-running circadian activity rhythms(Rosenwasser et al., 2005). Another study in rat also found thatchronic EtOH consumption impaired the circadian rhythm ofproopiomelanocortin (POMC) and the clock-governing rat periodgenes mRNA expression in the hypothalamus (Chen et al., 2004).These pilot studies support our results about sleep disorder causedby acute or chronic administration of 0.1% EtOH. Furthermore,in this experiment we found that acute exposure of 0.1% EtOHwas more severe than the chronic one during the dark cycle.This phenomenon may be caused by an adaptation ability ofzebrafish to EtOH. This statement is supported by another studythat provides evidence of adaptation in adult zebrafish to EtOH,showing that after a 2-week long chronic exposure to EtOH,treated zebrafish behaved the same way as the fish that were nevertreated with EtOH (Gerlai et al., 2006). Afterwards, we cross-examined the locomotion activity patterns for both EtOH exposuregroups, analyzed by idTracker and ImageJ methods. Statisticalanalyses were conducted after data normalization and later, wefound that that both sets of data were parallel and correlated eachother. The r-values from Spearman’s nonparametric correlationtest also showed that the two variables (ImageJ and idTrackerresults) correlate with each other (Fig. S3C,D for acute 0.1%EtOH and Fig. S3E,F for chronic 0.1% EtOH).

Comparison of instrument settings between previous andcurrent study for circadian rhythm measurementIn rodents, the circadian rhythm can be measured by counting howoften they interrupt the infrared (IR) light (Clarke et al., 1985) ormicrowave signals (Genewsky et al., 2017) in the mouse cage. Incatfish, a similar IR-based method has been utilized to analyzecatfish circadian rhythm and has found that the catfish is a nocturnalspecies (Ramteke et al., 2009). The sensitivity of the IR sensor-based method largely relies on the number of sensors; movementevents are counted once if the IR signals are interrupted. The IR-sensor method is unable to measure detailed parameters like movingvelocity, turning angle and meandering. In this study, by using anidTracker-based approach, we were able to measure these diverselocomotion parameters for circadian rhythm analysis in fish for thefirst time.

For the zebrafish research community, the most popular methodto conduct circadian rhythm locomotion activity measurement isbased on using commercial instruments and software (likeZebraCube from Viewpoint Company, http://www.viewpoint.fr/en/p/equipment/zebracube). However, the cost of such instrumentsand third-party software is usually unaffordable to most researchers.This disadvantage encouraged us to assemble a low-cost instrumentcoupled with the two open-source software, idTracker and ImageJ,to perform circadian rhythm studies in fish. The total cost for theentire setup is estimated to be under 3000 USD, while thecommercial instrument usually costs around 30,000 USD. Inaddition, due to growing interest within the zebrafish community,the ImageJ-based method for zebrafish behavioral analyses isgaining popularity in this field. However, most of these studies wereconducted mostly in zebrafish larvae, while studies in adultzebrafish are quite rare (Nema et al., 2016). Moreover, most ofthese methods have been focused on other zebrafish behaviors, suchas avoidance behavior (Pelkowski et al., 2011), anxiety-relatedbehavior (Richendrfer et al., 2012) and locomotion behavior(Colwill and Creton, 2011). Previous circadian rhythm-relatedstudies have been focused on larvae fish, thus, the circadian rhythmfor adult fish is seldom investigated. In zebrafish, the circadianrhythm patterns in larvae and adults are distinct. Larvae show highactivity during the dark cycle (scotophase) and sleep-like behaviorduring the light cycle (photophase) (Padilla et al., 2011). On theother hand, adult zebrafish display sleep-like behavior during thedark cycle and high activity during the light cycle. We believe oursimple, cost-effective device can provide a useful platform to screengenetic mutants in adult zebrafish with circadian rhythm deficiencywith high resolution and reliability.

DISCUSSIONComparison of performance between idTracker and ImageJfor circadian rhythm measurementIn this report, we demonstrated how two image-based methods,idTracker and ImageJ, can be used to calculate the circadianrhythm locomotion activity for both diurnal (zebrafish) andnocturnal (catfish) fish species. After cross-validation, bothmethods can provide consistent and reliable data for circadianrhythm locomotion analysis (Fig. 2B–G, Figs S1A–C, S2 and S3).The pros and cons of each method are compared and summarizedin Table 1. The most significant advantage for the ImageJ-basedmethod is a faster calculation speed by running a macro script.Although the detailed parameters such as swimming speed,angular velocity and meandering could not be obtained, theImageJ-based method provides an excellent tool for quickscreening and evaluation. For example, it only takes 3 min ofanalysis time for videos with 18 fish (six tanks with three fish foreach tank) using the ImageJ-based method. However, it will takearound 13 min to finish data analysis using the idTracker-basedmethod. Therefore, we recommend the ImageJ-based method as aninitial screening platform. Once some interesting differences incircadian rhythm pattern were detected using this fast and simpleImageJ-based method, we could then apply the idTracker-basedmethod to extract more detailed locomotion parameters such asswimming speed, angular velocity and meandering. However, onemust keep in mind before using this ImageJ-based method that thismethod is very sensitive towards background noise, such as dark-line artefacts induced by camera-acquisition speed with respect tothe LED lights. Therefore, noise removal is a prerequisite beforeconducting this experiment, as noise can disrupt the analysis, asmentioned in Table 1.

6

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

In conclusion, we have established a cost-effective and easy setupdevice for circadian rhythm analysis, and also validated twoalgorithms (idTracker and ImageJ) suitable to perform circadianrhythm locomotion measurement for both diurnal (zebrafish) andnocturnal (catfish) fish species for the first time. The methodreported here can be used to evaluate the potential toxicity ofchemicals on fish, perform psychological drug screenings onregulating the circadian rhythm and screen genetic mutationsdisplaying circadian rhythm impairment.

MATERIALS AND METHODSAnimal ethicsAll the experimental protocols and procedures involving zebrafish wereapproved by the Committee for Animal Experimentation of the Chung YuanChristian University (Number: CYCU104024, issue date 21 December2015). All experiments were performed in accordance with the guidelinesfor laboratory animals. Zebrafish (AB strain) were obtained from TaiwanZebrafish Core Facility at Academia Sinica, Taipei, Taiwan, and catfish(C. batrachus) were purchased from a local aquarium. Catfish weretransferred to the Zebrafish Core Facility and acclimatized to the core facilityphoto regime for ∼1 month. The Zebrafish Core Facility maintained a 12:12photo regime. Acclimation was necessary to habituate both of the fish in thecurrent photo period condition with white-color LED light as substitute fornatural light during the light cycle. Natural daylight has color temperaturesranging from 6000 K under overcast conditions to as high as 20,000 K underclear blue sky conditions during midday periods. However, in the eveninghours, the color temperature of the sun decreases to only 2000 K.Meanwhile, common LED lights have stable color temperatures rangingfrom 4000 K–4500 K (Schubert and Kim, 2005).

Zebrafish exposed to EtOHFor the acute toxicity test, adult zebrafish aged 6 months were exposed toEtOH for either 30 min (acute) or 7 days (chronic) at a low dosage of 0.1%concentration. About 90% of the water was changed every 24 h withredosing after each change. After EtOH exposure, fish were moved into thecircadian rhythm measuring chamber for filming and locomotion tracking.

Circadian rhythm chamber constructionSleep–wake behaviors were evaluated using a custom-designed circadianrhythm chamber (designed by Zgenebio Inc., http://www.zgenebio.com/).The light–dark cycle test apparatus consisted of six small plastic fish tanks(20×10 cm) which were placed above a custom-made light box containing

IR LEDs and white-color LED stripes arrayed side-by-side. For the lightcycle, white-color LEDs were turned on, triggered by timer 1. For the darkcycle, IR LEDs were turned on, triggered by timer 2. An IR-sensitive CCDcamerawith magnifying lens without image distortion was located above theexperimental setup to record the fish movements at 30 frames per second(fps). To maintain video frame rate, the IR CCD camera was connected to aworkstation computer with a 1 TB SSD (PLEXTOR M9PeY PCIe, Taipei,Taiwan) and advanced CPU in order to process the data writing and readingefficiently. In addition, Total Recorder software (http://www.totalrecorder.com) was used to perform schedule recording. The whole assembly wasplaced inside a thermal incubator and the regular temperature for circadianrhythm testing was set at 25°C.

Video tracking and data analysisThe locomotion activity in the videos was analyzed by using open-sourcesoftware. For the idTracker-based method (Audira et al., 2018; Pérez-Escudero et al., 2014), the X and Y coordinates of the tracked animals wereextracted from the videos frame-by-frame and analyzed following themethod described in the Supplementary information (see protocol). For theImageJ-based method (Abràmoff et al., 2004), the dynamic pixel-changemethod was employed for locomotion activity extraction. The detailedprotocol for data analysis can be found in the Supplementary information(see protocol). All tests were analyzed by either the parametric test (unpairedt-test or Dunnett’s multiple comparisons test) or the non-parametric test(Mann–Whitney test) based on the data group for each experiment. Due tothe differences between the idTracker and ImageJ result unit, datanormalization is required in order to compare the light and dark cycledata patterns. Data normalization was done using GraphPad Prism software–with 0% defined as the smallest value in each data set and 100% defined aslargest value in each data set – and analyzed with an unpaired t-test withWelch’s correction, since we cannot assume that these two populations haveequal standard deviations. Later, a correlation test was also conducted inorder to validate the equivalence of the two methods’ results. As every dataset consisted of both of light and dark cycles results, Spearman’s test – a testthat makes no assumption about the distribution of the values – was used.Using this nonparametric correlation test, the correlation coefficient (r) canbe obtained. If the value of r is between −1 and 1, it means that these twovariables tend to negatively or positively associate together, while an r-valueof 0 means that the two variables do not correlate at all.

Construction of a cost-effective light box chamber for circadianrhythm studyIn this study, we aimed to establish an easy-to-set-up device that consisted ofa custom-made light box chamber equipped with white 940 nm LED lightstrips as light sources suitable for studying circadian rhythm of diurnal (i.e.zebrafish) and nocturnal (i.e. catfish) fish species with reliability andreproducibility. Six small acrylic fish tanks (20×10 cm) were placed above alight box (Fig. 1A). The light box (48×48 cm) contained 12 strips of whitechip on board (COB) LEDs (each strip contained 15 COB LED bubbles) asthe light source in the light cycle (Fig. 1B), and 10 strips of 940 nm IR LEDs(each strip contained 15 LED bubbles) as the light source in the dark cycle(Fig. 1C). They were arranged side-by-side as shown in Fig. 1D. The timers,which connected between the light box and power source, were used tocontrol the timing for the light cycle (day cycle, COB LED on) and darkcycle (night cycle, 940 nm IR LED on). To reduce video distortion, a940 nm IR CCD camera with a magnifying lens was used to record fishmovements at 30 fps. Using this device, we were able to track the zebrafishlocomotion during the light and dark cycles with high-resolution outputvideo (up to 1028×1024 pixels) (Fig. 1B,C). The light/dark cycle was set as12:12 photo-regimen. Next, we recorded the fish locomotion activity for1 min at 1 h intervals and then used either idTracker (Pérez-Escudero et al.,2014) or ImageJ (Abràmoff et al., 2004) software to analyze locomotionactivities. The video format was converted from compressed AVI touncompressed AVI by VirtualDub software (https://sourceforge.net/projects/virtualdub/) in order to be compatible with ImageJ software(Nema et al., 2016). Such video format conversion is unnecessary foridTracker software. Finally, the relative locomotion activities obtained fromboth algorithms were compared using statistical software GraphPad Prism.

Table 1. Comparison of idTracker- and ImageJ-based methods incalculating circadian rhythm

idTracker ImageJ

Programming language-based

Matlab Java

Endpoint output Morerepresentative

Only in pixel intensitychanges

Occluded fish analysis Hard to analyze Easy to analyzeBuilt-in macro forautomatization

Not available Available

Analysis time needed(6 tanks, 1 fish)

5′35″ 3′ (without macro),2′ (with macro)

Analysis time needed(6 tanks, 3 fish)

13′ 3′ (without macro),2′ (with macro)

Video format supported Many formats Only AVIFish-size dependent No YesIndividual fish tracking Able Not ableDead fish/noise Can be excluded Possible to be countedData storage consumptionsize (24 videos)

Smaller(∼15.5 Gb)

Bigger (∼349 Gb)

Trajectories Able Not ableRAM consumption 10 Gb 4 GbCPU Usage 75% @3.3 GHz 10% @3.3 GHz

7

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

Measurement of the locomotion activity of fish using idTrackeror ImageJHere, we provide two simple and easy protocols to perform circadian rhythmactivity measurement. The schematic in Fig. 2A summarizes the entireanalysis pipeline. The first is an idTracker-based method that can be used totrack locomotion at single-individual resolution. The second protocol is anImageJ-based method that can be used to measure fish locomotion in a quickand simple manner by using macro-language script (detailed protocolsfor both methods are attached in the Supplementary information underSupplementary protocol). For the idTracker-based protocol, the X and Ycoordinates can be extracted from videos for each individual, making itpossible to measure the fish locomotion activity more precisely. idTracker isalso able to obtain some important parameters, such as swimming speed,swimming distance, turning angle, angular velocity, meandering, freezing,swimming and rapid time–movement ratio. After every parameter for eachtime interval obtained, we grouped the parameters by their cycle andconducted statistical analysis. Furthermore, for the ImageJ-based protocol, thedynamic pixel changes of adjacent images were extracted frame-by-frame,making it possible to qualitatively measure the overall fish locomotion activityin a more robust manner with the aid of macro language. However, thequantitative information such as swimming speed, swimming distance,turning angle, angular velocity, meandering, freezing, swimming and rapidtime–movement ratio are unobtainablewhen using the ImageJ-based protocol.

AcknowledgementsWe appreciate the professional comments from two antonymous reviewers toenhance our paper quality.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: Y.-H.L., C.-D.H.; Methodology: G.A., B.P.S., C.-D.H.; Validation:G.A., B.P.S., S.J., S.-T.L., C.-D.H.; Formal analysis: G.A., S.-T.L., C.-D.H.;Investigation: L.H., C.-D.H.; Resources: L.H., C.-D.H.; Data curation: B.P.S., S.J.,S.-T.L., C.-D.H.; Writing - original draft: G.A., C.-D.H.; Visualization: C.-D.H.;Supervision: Y.-H.L., L.H., C.-D.H.; Project administration: C.-D.H.; Fundingacquisition: C.-D.H.

FundingThis research received funding from theMinistry of Science and Technology, Taiwan(grant no. 107-2622-B-033-001-CC2).

Supplementary informationSupplementary information available online athttp://bio.biologists.org/lookup/doi/10.1242/bio.041871.supplemental

ReferencesAbramoff, M. D., Magalhaes, P. J. and Ram, S. J. (2004). Image processing withImageJ. Biophotonics. Int. 11, 36-42.

Antoch, M. P., Song, E.-J., Chang, A.-M., Vitaterna, M. H., Zhao, Y.,Wilsbacher, L. D., Sangoram, A. M., King, D. P., Pinto, L. H. and Takahashi,J. S. (1997). Functional identification of the mouse circadian Clock geneby transgenic BAC rescue.Cell 89, 655-667. doi:10.1016/S0092-8674(00) 80246-9

Audira, G., Sampurna, B. P., Juniardi, S., Liang, S.-T., Lai, Y.-H. andHsiao, C.-D.(2018). A simple setup to perform 3D locomotion tracking in zebrafish by using asingle camera. Inventions 3, 11. doi:10.3390/inventions3010011

Cahill, G. M., Hurd, M.W. and Batchelor, M. M. (1998). Circadian rhythmicity in thelocomotor activity of larval zebrafish. Neuroreport 9, 3445-3449. doi:10.1097/00001756-199810260-00020

Chen, C. P., Kuhn, P., Advis, J. P. and Sarkar, D. K. (2004). Chronic ethanolconsumption impairs the circadian rhythm of pro-opiomelanocortin and periodgenes mRNA expression in the hypothalamus of the male rat. J. Neurochem. 88,1547-1554. doi:10.1046/j.1471-4159.2003.02300.x

Clarke, R. L., Smith, R. F. and Justesen, D. R. (1985). An infrared device fordetecting locomotor activity. Behav. Res. Methods Instruments Comput. 17,519-525. doi:10.3758/BF03207645

Colwill, R. M. andCreton, R. (2011). Locomotor behaviors in zebrafish (Danio rerio)larvae. Behav. Process 86, 222-229. doi:10.1016/j.beproc.2010.12.003

Delaunay, F., Thisse, C., Marchand, O., Laudet, V. and Thisse, B. (2000). Aninherited functional circadian clock in zebrafish embryos. Science 289, 297-300.doi:10.1126/science.289.5477.297

Dijk, D.-J., Duffy, J. F. and Czeisler, C. A. (2001). Age-related increase inawakenings: impaired consolidation of nonREM sleep at all circadian phases.Sleep 24, 565-577. doi:10.1093/sleep/24.5.565

Duffy, J. F., Zitting, K.-M. and Chinoy, E. D. (2015). Aging and circadian rhythms.Sleep Med. Clin. 10, 423-434. doi:10.1016/j.jsmc.2015.08.002

Ehlers, C. L. and Kupfer, D. J. (1989). Effects of age on delta and REM sleepparameters. Electroencephalogr. Clin. Neurophysiol. 72, 118-125. doi:10.1016/0013-4694(89)90172-7

Gamse, J. T., Shen, Y.-C., Thisse, C., Thisse, B., Raymond, P. A., Halpern,M. E. and Liang, J. O. (2002). Otx5 regulates genes that show circadianexpression in the zebrafish pineal complex.Nat.Genet. 30, 117. doi:10.1038/ng793

Gandhi, A. V., Mosser, E. A., Oikonomou, G. and Prober, D. A. (2015). Melatoninis required for the circadian regulation of sleep. Neuron 85, 1193-1199. doi:10.1016/j.neuron.2015.02.016

Genewsky, A., Heinz, D. E., Kaplick, P.M., Kilonzo, K. andWotjak, C. T. (2017). Asimplified microwave-based motion detector for home cage activity monitoring inmice. J. Biol. Eng. 11, 36. doi:10.1186/s13036-017-0079-y

Gerlai, R., Lahav, M., Guo, S. andRosenthal, A. (2000). Drinks like a fish: zebra fish(Danio rerio) as a behavior genetic model to study alcohol effects. Pharmacol.Biochem. Behav. 67, 773-782. doi:10.1016/S0091-3057(00)00422-6

Gerlai, R., Lee, V. and Blaser, R. (2006). Effects of acute and chronic ethanolexposure on the behavior of adult zebrafish (Danio rerio). Pharmacol. Biochem.Behav. 85, 752-761. doi:10.1016/j.pbb.2006.11.010

Gilbert, M. J. H., Zerulla, T. C. andTierney, K. B. (2014). Zebrafish (Danio rerio) as amodel for the study of aging and exercise: physical ability and trainability decreasewith age. Exp. Gerontol. 50, 106-113. doi:10.1016/j.exger.2013.11.013

Hirayama, J., Cardone, L., Doi, M. and Sassone-Corsi, P. (2005). Commonpathways in circadian and cell cycle clocks: light-dependent activation of Fos/AP-1 in zebrafish controls CRY-1a and WEE-1. Proc. Natl Acad. Sci. USA 102,10194-10199. doi:10.1073/pnas.0502610102

Hurd, M.W. andCahill, G. M. (2002). Entraining signals initiate behavioral circadianrhythmicity in larval zebrafish. J. Biol. Rhythms 17, 307-314. doi:10.1177/074873002129002618

Hurd, M. W., Debruyne, J., Straume, M. and Cahill, G. M. (1998). Circadianrhythms of locomotor activity in zebrafish. Physiol. Behav. 65, 465-472. doi:10.1016/S0031-9384(98)00183-8

Kasai, M., Yamamoto, T. and Kiyohara, S. (2009). Circadian locomotor activity inJapanese sea catfish Plotosus lineatus.Fish. Sci. 75, 81-89. doi:10.1007/s12562-008-0001-0

King, D. P., Zhao, Y., Sangoram, A. M., Wilsbacher, L. D., Tanaka, M., Antoch,M. P., Steeves, T. D., Vitaterna, M. H., Kornhauser, J. M., Lowrey, P. L. et al.(1997). Positional cloning of the mouse circadian clock gene. Cell 89, 641-653.doi:10.1016/S0092-8674(00)80245-7

Konopka, R. J. and Benzer, S. (1971). Clock mutants of Drosophila melanogaster.Proc. Natl Acad. Sci. USA 68, 2112-2116. doi:10.1073/pnas.68.9.2112

Lahiri, K., Vallone, D., Gondi, S. B., Santoriello, C., Dickmeis, T. and Foulkes,N. S. (2005). Temperature regulates transcription in the zebrafish circadian clock.PLoS Biol. 3, e351. doi:10.1371/journal.pbio.0030351

Lockwood, B., Bjerke, S., Kobayashi, K. and Guo, S. (2004). Acute effects ofalcohol on larval zebrafish: a genetic system for large-scale screening.Pharmacol. Biochem. Behav. 77, 647-654. doi:10.1016/j.pbb.2004.01.003

Lopez-Olmeda, J. F. and Sanchez-Vazquez, F. J. (2011). Thermal biology ofzebrafish (Danio rerio). J. Therm. Biol. 36, 91-104. doi:10.1016/j.jtherbio.2010.12.005

Lopez-Olmeda, J. F., Madrid, J. A. and Sanchez-Vazquez, F. J. (2006). Light andtemperature cycles as zeitgebers of zebrafish (Danio rerio) circadian activityrhythms. Chronobiol. Int. 23, 537-550. doi:10.1080/07420520600651065

McClelland, G. B., Craig, P. M., Dhekney, K. and Dipardo, S. (2006).Temperature- and exercise-induced gene expression and metabolic enzymechanges in skeletal muscle of adult zebrafish (Danio rerio). J. Physiol. 577,739-751. doi:10.1113/jphysiol.2006.119032

Nema, S., Hasan, W., Bhargava, A. and Bhargava, Y. (2016). A novel method forautomated tracking and quantification of adult zebrafish behaviour during anxiety.J. Neurosci. Methods 271, 65-75. doi:10.1016/j.jneumeth.2016.07.004

Padilla, S., Hunter, D. L., Padnos, B., Frady, S. and MacPhail, R. C. (2011).Assessing locomotor activity in larval zebrafish: Influence of extrinsic and intrinsicvariables. Neurotoxicol. Teratol. 33, 624-630. doi:10.1016/j.ntt.2011.08.005

Pelkowski, S. D., Kapoor, M., Richendrfer, H. A., Wang, X., Colwill, R. M. andCreton, R. (2011). A novel high-throughput imaging system for automatedanalyses of avoidance behavior in zebrafish larvae. Behav. Brain Res. 223,135-144. doi:10.1016/j.bbr.2011.04.033

Perez-Escudero, A., Vicente-Page, J., Hinz, R. C., Arganda, S. and DePolavieja, G. G. (2014). idTracker: tracking individuals in a group by automaticidentification of unmarked animals.Nat.Methods11, 743. doi:10.1038/nmeth.2994

Ralph, M. R. and Menaker, M. (1988). A mutation of the circadian system in goldenhamsters. Science 241, 1225-1227. doi:10.1126/science.3413487

Ramteke, A. K., Poddar, P. and Pati, A. K. (2009). Circadian rhythms of locomotoractivity in Indian walking catfish, Clarias batrachus. Biol. Rhythm. Res. 40,201-209. doi:10.1080/09291010802069169

Richendrfer, H., Pelkowski, S. D., Colwill, R. M. and Creton, R. (2012). Onthe edge: pharmacological evidence for anxiety-related behavior in zebrafishlarvae. Behav. Brain Res. 228, 99-106. doi:10.1016/j.bbr.2011.11.041

8

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from

Rosenwasser, A. M., Fecteau, M. E. and Logan, R. W. (2005). Effects ofethanol intake and ethanol withdrawal on free-running circadian activity rhythms inrats. Physiol. Behav. 84, 537-542. doi:10.1016/j.physbeh.2005.01.016

Schubert, E. F. and Kim, J. K. J. S. (2005). Solid-state light sources getting smart.Science 308, 1274-1278. doi:10.1126/science.1108712

Tamai, T. K., Young, L. C. and Whitmore, D. (2007). Light signaling to thezebrafish circadian clock by Cryptochrome 1a. Proc. Natl Acad. Sci. USA 104,14712-14717. doi:10.1073/pnas.0704588104

Vatine, G., Vallone, D., Gothilf, Y. and Foulkes, N. S. (2011). It’s timeto swim! Zebrafish and the circadian clock. FEBS Lett. 585, 1485-1494. doi:10.1016/j.febslet.2011.04.007

Vera, L. M., Al-Khamees, S. and Herve, M. (2011). Stocking density affectscircadian rhythms of locomotor activity in African catfish, Clarias gariepinus.Chronobiol. Int. 28, 751-757. doi:10.3109/07420528.2011.606388

Webb, W. B. (1982). The measurement and characteristics of sleep inolder personsv of sleep in older persons. Neurobiol. Aging 3, 311-319. doi:10.1016/0197-4580(82)90019-7

Zhdanova, I., Yu, L., Lopez-Patino, M., Shang, E., Kishi, S. andGuelin, E. (2008).Aging of the circadian system in zebrafish and the effects of melatonin onsleep and cognitive performance. Brain Res. Bull. 75, 433-441. doi:10.1016/j.brainresbull.2007.10.053

9

METHODS & TECHNIQUES Biology Open (2019) 8, bio041871. doi:10.1242/bio.041871

BiologyOpen

by guest on March 18, 2021http://bio.biologists.org/Downloaded from